singularityhacker/estack/estack-quality
Claude Code and agent skill by singularityhacker/estack
- community
Checking that the SKILL.md file is still in the repository.
Configure and run an Elixir/Phoenix quality gate with honest failure handling. Use for mix verify/precommit, Credo, coverage, Dialyzer, Sobelow, dependency audits, or CI configuration.
Install
npx vskill@latest install singularityhacker/estack/estack-qualitySKILL.md
The quality wall
Inspect the existing aliases, dependency versions, and CI first. Preserve working gates. Use the project's approved tooling; do not introduce every optional dependency just to match this reference.
Start with an executable core
The following goes inside the application's Mix project module. Merge it with existing aliases and project settings rather than replacing them:
Run the core with MIX_ENV=test mix verify; develop with focused tests. Before publication, run the full approved gate. Never claim a check ran because its dependency appears in mix.lock.
Add checks intentionally
| Check | Prerequisite / configuration | |
|---|---|---|
credo --strict | Credo dependency and valid .credo.exs | |
coveralls.json --raise | ExCoveralls, test_coverage, test environment, and a configured threshold | |
dialyzer | Dialyxir plus appropriate PLT caching | |
sobelow --config | Phoenix app, Sobelow, and a reviewed .sobelow-conf | |
deps.audit | mix_audit with a current advisory database | |
hex.audit | Hex retired-package check; distinct from vulnerability auditing | |
excellent_migrations.check_safety | Compatible migration checker, database assumptions, and reviewed configuration | |
deps.unlock --check-unused | Mix version supporting that flag |
Use coverage in place of the ordinary full test task if it already runs the same suite, rather than running it twice. Declare the necessary dependencies before adding commands to an alias. There are no required ClawBank-specific Mix tasks.
Verify every selected command's options with mix help and the installed version. Configure security exit thresholds deliberately: a stricter threshold must not be described as weakening the gate. Ensure the CI job propagates nonzero exit status.
Bundled custom check
Copy discarded_result.ex to lib/mix/estack/discarded_result.ex only if adopting the check. It is guarded so production builds without the dev/test Credo dependency still compile. Merge the relevant configuration from credo.exs.example.
The check flags explicit _ = assignments to selected Repo writes, Oban.insert, and Phoenix.PubSub.broadcast, including a final piped call. Handle expected failures with case/with or an intentional match. It does not prove all results are handled, infer arbitrary aliases, or flag every unused expression. Configure repo_modules for aliases such as DB. It deliberately does not flag _ = Ecto.Multi.new() as a discarded error: that constructor does not return an error tuple.
Coverage
Use coveralls.json.example as a starting configuration, then choose a project-appropriate minimum_coverage. floor_coverage controls rounding only. Do not claim per-file thresholds unless the installed tool/config implements them.
Test important authorization, state transitions, errors, and retries regardless of the overall percentage. Establish a realistic baseline for existing code and improve it deliberately.
Preserve integrity
Fix the finding rather than deleting a check. Explain narrow, reviewed exceptions next to the configuration or code. Legitimate gate repairs are allowed when authorized; avoid adding ignores or exclusions solely to make CI green.
Use the existing toolchain pin (.tool-versions, mise, or the project's equivalent) consistently across CI and deployment. Respect formatter/line-length conventions. Cache dependency/PLT artifacts using keys that include relevant toolchain and lockfile changes.
References
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/estack-quality/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/singularityhacker/estack/estack-quality)